I don't think there are any predefined names for the shape of the spectrogram. If there are, these are unlikely to be very useful as the shape will differ between individuals, populations and/or species.

The fundamental frequency is the clearest marker in the syllable in your spectrogram. This is often the bottom one in a harmonic call, but not necessarily so. In your case I'd say b2 is the fundamental (or on the left half of your spectrogram, the second line from the bottom).

I would agree that this call is non-tonal. First of all your call consists of several shorter pulses and for each of these single pulses, the upper sidebands are not multiples of the fundamentals (which would be the case for tonal sounds). Another indicator for the non-tonal character of that call (or each pulse) is that sidebands with most energy (e.g. your b2) are not the fundamental (your F0). I am not sure whether your call is recorded on-axis throughout the whole call. For many vocs, the frequency content changes with changing recording axis which makes it more complicate. I would agree that F0 is the fundamental. And when the second part is recorded on-axis as well, there is a sudden shift of the fundamental to b2.

If there is only one animal calling there, it is possible that you have a kind of nonlinear phenomena common in mammalian vocalizations (biphonation). So, you probably have two independent fundamental frequencies (as you can see in the vocal units from the middle of the call on). It seems that at the begining of the sequence the animal is producing tonal calls that the fundamental is the F0. Then you have the biphonation happening.

Please, find more about biphonation in these papers:

Riede, T., Herzel, H., Mehwald, D., Seidner, W., Trumler, E., Tembrock, G. & Bo¨hme, G. 2000: Nonlinear phenomena and their anatomical basis in the natural howling of a female dog-wolf breed. J. Acoust. Soc. Am. 108, 1435—1442

Fitch, W. T., Neubauer, J. & Herzel, H. 2002: Calls out of chaos: the adaptive significance of nonlinear phenomena in mammalian vocal production. Anim. Behav. 63, 407—418.

Thank you everyone for the suggestions, they are very helpful.

The spectrogram is a loud call from a single individual adult male. And all the loudcall recordings (from various adult male individuals) have the same pattern. The left side of the dashed line has this combination of type a, b1, and b2. On the right side of the dashed line, type a disappears leaving only b1 and b2.

I would suggest to be very careful while describing this loud call, both regarding the biphonation and the tonal vs atonal quality of the call.

It looks like you can't be categorical here, since the call structure differs with time, starting with short (you named them 'a'), relatively tonal calls (0 to 0.8s, I would say F0 is the fundamental at this stage). Then the structure of these short calls gets less tonal, while other "notes" or "syllables" appear(you named them 'b1' and 'b2' (P.S: you could name these different call, since usually the terms 'note' and 'syllable' apply more to bird songs). 'b1' and 'b2' do not seem to be a case of biphonation, since they do not seem to be alined in time ('b1' produced before 'b2'; then in this context, you would have a very high frequency modulation, with fundamental frequency following 'b1', then 'b2').

The only question I would potentially have regarding biphonation at this point would be 'a' vs 'b2': a seems to fade away while 'b1/b2' appear. I would suggest to modify the 'spectrogram view' parameters to have a closer (focus on the frequency range of interest) and sharper (play with the window size settings to get frequencies sharply delimited in time) look at the various "notes". Also make sure that only one individual is producing the signal.

Interesting case anyways!

I would say that first part are very short tonal sounds and I agree that fundamental would be F0.

I also think that there is a biphonation phenomenon on the recording, but not thinking in b1 and b2 as result of a different phonic system, agree with Maxime that these can be produced by the same phonic system, as they are not simultaneous.

I would suggest that the first part of the call is being produced by one phonic system, oscilating at F0 and between units 6 and 4 the second phonic system would start producing the b1 and b2 signal. During the transition phase (from unit 6-4 to unit 1) the signal shows mixed characteristics of both systems, and finally the first phonic system stops vibrating and  the second part shows the free vibration of the second phonic system.

It would be nice to have some info on the parameters you used to calculate the spectrogram (size of window and overlapping can be critical to get more detail either in the time or frequency domains). 

Dear Larissa,

 I would concur that your estimate of F0 for the early units (12 to 6) is probably correct, although reading it off of a spectrogram of this resolution would be difficult, and you might want to consider looking for cycles in the time domain.  Provided that the vocal fold periodicity is not changing too rapidly, it is usually more accurate to find the duration of several cycles, and divide their average duration into a second as the measurement error is distributed over the multiple cycles.  This would also let you readily see if the areas where you see two apparent sources are caused by alternating glottal cycles of different length, although there are other causes that would not be readily evident and others have suggested articles for this.     

 If you are interested in trying to automate your estimates, many people use the freeware programs wavesurfer (http://www.speech.kth.se/wavesurfer/) and Praat (http://www.fon.hum.uva.nl/praat/).  Personally, I use wavesurfer although I have found the occasional bug, but Praat is also very good and I haven’t looked at the recent versions.  Regardless, be sure to do a sanity check on the outputs as F0 estimators are notorious for occasionally doubling the frequency.   

 Coming from a speech background before moving to bioacoustics, I prefer thinking about this as a voiced call than a tonal call; voicing simply implies that the vocal folds are in vibration and F0 is the fundamental frequency at which they are vibrating.  Harmonics and formant structures can usually be found in voiced calls (Tecumseh Fitch has a very nice review article that includes a tutorial on formant structure – 2000, Trends Cog Sci 4(7) pp 258-67), but they need not necessarily be tonal.  Constrictions in the vocal tract can introduce turbulence for voiced as well as unvoiced calls which can frequently be seen as high frequency noise, such as one might here in the /f/ or /v/ phones produced by humans (the former is unvoiced, the latter is voiced).

Best regards,

Marie

I would suggest to investigate the possibility that - anyhow the first part of - the recording consists of pulsed vocalizations with a fundamental frequency (within the individual pulses) of approximately 1.5 kHz, in which the pulse repetition frequency (PRF) is superimposed on the fundamental. See for an explanation my paper "Acoustic signals by harbour porpoises (Phocoena phocoena)" (via Researchgate). Maybe also my paper "Spectral analysis of vocalisations of the Moluccan megapode Eulipoa wallacei (Aves: Megapodiidae) during egg-laying activities" is interesting. To find out, carry out an accurate FFT spectral analysis (sound pressure versus frequency analysis).

All the best,

Wim

Dear all,

thank you for the suggestions. Attached are spectrograms with different parameters and zoomed in. The recording was sampled at 48kHz/24bit. I'm using Raven Pro for my acoustic analysis, which does not have automatic measurement for F0.

Hi Larissa. Thank you for providing more details. It isn’t an easy sound to interpret. Try to gain more resolution in time (short window size) to see if you are able to understand where each element starts and end so you can discuss if you have biphonation happening there or not. In the last spectrogram you provided us, it seems that you have the element “a” with fundamental F0, then the animal starts to produce other elements before “a” (“b1” followed by “b2” followed by “a”). For me it seems that the element “a” do not overlap with the others, but I’m not sure whether b1 and b2 are overlapping in time. If they are, maybe you have two independent fundamental frequencies in a sound spectrum (biphonation). I faced something similar in titi monkey’s calls. If you want, I can send you a sample. In my case I decide to consider the elements produced very close as a unit since it was always emitted together.

Dear Christini,

Thank you very much for the suggestions. I have inspected closer to each unit and element "b1" and "b2" does overlap in time. But "b2" does not overlap with "a". Which means there is a case of biphonation between b1 and b2. So does that mean each element (a, b1, b2) have its own fundamental frequency? It'd be great if you don't mind sending me a sample so I can compare it with my case. Thank you.

When you look at the spectrograms, you see that the highest levels are concentrated around 1.5 kHz. This may be caused by: 1) the measurement system does not have a flat response, but a peak around 1.5 kHz; 2) the vocalization is a pulsed signal with a fundamental of 1.5 kHz on which the PRF is superimposed. Most probable is the second cause. You should do a FFT and present the result as sound pressure - frequency diagram. If there is a prominent peak ('mountain') around 1.5 kHz, with small peaks at both sides, reason 2) is the case.  If you are interested, you may send me the recording, so I will look at it.                                 Best,  Wim

These are clear examples of biphonation in Killer whales, Not really similar to your sound, but I hope this can help you understand the concept. You can clearly see two independent vibrations simultaneosuly, with two different fundamental frequencies, and evolving independently in time.

Just be careful to distinguish between 'biphonic' (one sound-producing structure generating multiple frequencies) and 'two-voiced' (two sound-producing structures producing two independent frequencies). Killer whales are likely an example of the latter (two nasal plugs operating independently). The following paper offers more detail: 

Wilden, I., Herzel, H., Peters , G. & Tembrock, G. 1998. Subharmonics, biphonation and deterministic chaos in mammal vocalizations. Bioacoustics, 9, 171-196.